A neural knowledge graph for codebases — three-layer code/bugs/tasks understanding for Claude Code
Wellknown found it in public sources; nobody has proven control of it yet. Claiming takes one click if the repository is under your GitHub account, or a small file on your domain otherwise. Verified owners get the badge, 15-minute checks, status alerts, edits that outrank crawled data, and a ranking boost.
Agents can do it too: POST https://wellknown.network/api/v1/claims with {"agent":"yakoub-ai-neural-memory-mcp","method":"well_known_file"} — machine-readable steps at claim.json, guide at /docs/claim.
Everything here was measured by our prober or read from a registry. Nothing is self-reported.
Attributed to the source that supplied each field. Treated as claims, not facts.
# Neural Memory [](LICENSE) [](https://www.python.org/downloads/) [](https://github.com/Yakoub-ai/neural-memory) > A persistent knowledge graph plugin for Claude Code that maps your codebase into four layers — code, bugs, tasks, and insights — so Claude always has deep context without burning tokens re-reading files.  --- ## What It Does Neural Memory builds and maintains a **four-layer directed knowledge graph** of your project: | Layer | Nodes | Purpose | |-------|-------|---------| | **Codebase** | modules, classes, functions, methods, overviews | Understands structure, calls, imports, inheritance | | **Bugs** | bug entries with severity and status | Tracks known issues linked to code nodes | | **Tasks** | phases and tasks with lifecycle | Tracks work items linked to code + each other | | **Insights** | technical knowledge accumulated over sessions | Builds persistent project documentation | Every node gets: - **Layered summaries** — heuristic one-liner from AST, optionally enriched via Claude API - **138-dim composite embeddings** — 100-dim TF-IDF+SVD content vector + 38-dim structural graph features - **LSP enrichment** — type signatures and diagnostics from Pyright/pylsp - **Importance scores** — graph-centrality weighted by node type Search uses a **three-phase branch algorithm**: seed by cosine similarity → expand through graph edges with decay/pruning → re-rank by combined score. Sensitive values (API keys, tokens, passwords) are automatically redacted before storage. **Supports 8 languages**: Python, JavaScript/TypeScript, Go, Rust, Ruby, PHP, Java, C/C++. --- ## Quick Start ### Install via Claude Code plugin marketplace (recommended) ``` /plugin marketplace add Ya…
Mapped onto the structured taxonomy from declared text and observed tool names. Confidence shown for derived entries.
Every source is kept verbatim. Field changes are logged as events.